Year: 2025
Type: conference-paper
Abstract: Machine learning is a key component of modern data-driven technologies and artificial intelligence, enabling systems to learn from data and make informed decisions without explicit programming. This paper compares two fundamental approaches: supervised and unsupervised learning. In supervised learning, training data include both inputs and known outputs, allowing the model to learn accurate mappings for prediction and classification tasks. Proper construction of training, validation, and test sets is very important to ensure generalization and prevent overfitting. Supervised methods are fast, accurate, and widely applied in finance, healthcare, and image recognition. Unsupervised learning, by contrast, works with unlabeled data, identifying hidden patterns or structures without prior outcomes. It clusters data based on statistical properties using techniques such as k-means, hierarchical clustering, and principal component analysis, supporting applications such as pattern discovery and market segmentation. The paper outlines both methods’ principles, processes, advantages, limitations, and real-world applications.
Authors: Enkeleda Shehi, Roel Shehi
Institution: Tirana Business University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7204149471